Brent Sheather: Share beauty is only skin deep

Two weeks ago, we looked at a report by two credit analysts, Gary Jenkins and Jim Reid, based in the London office of Deutsche Bank, which illustrated why historic returns from United States bond and equity markets were not good indicators of future returns.

The fact that past returns are not a good guide to the future may come as a shock to some in the financial planning industry: much of the independent research on managed funds consists merely of calculating the return of each fund in each asset class and assigning star ratings to the top performers.

A few years back, one particularly simplistic analysis of historical data noted the outperformance of international equity funds relative to their local counterparts and recommended switching from New Zealand shares to international.

Jenkins and Reid's analysis illustrated that an asset class which performed well in the past often in the process becomes more expensive and thus is likely to underperform in the future, and vice versa.

With the benefit of hindsight, in December 2000, after four years of flat-lining, New Zealand shares were priced to outperform with low valuations and high dividends.

Jenkins and Reid then go on to suggest a more realistic way of estimating future returns, which is our focus this week.

Today, US shares boast an attractive long-term track record: 13.3 per cent a year over 25 years and 9.8 per cent a year since 1900.

Unfortunately, in the financial markets, beauty is often only skin deep and a high historic rate of return may simply set up an asset class to underperform in the future. So where to next for US shares?

This forecast requires two estimates, say the analysts: profits per share and the multiple which the stockmarket applies to these profits.

A central theme in the Jenkins-Reid analysis is mean-reversion. What this assumes is that valuations, earnings and interest rates oscillate around an average value sometimes higher, sometimes lower, but, ultimately, they revert back to their long-term average. The US sharemarket's average valuation between 1900 and 2005 is 15 times profits per share - however prices are at the high end of their range around 20 times profits, as shown in the graph.

To forecast what returns will be in the next five to 10 years, Jenkins and Reid assume that valuations will fall from the present above-average level of 20 times profits to the average of 15 times by 2010.

So we have the multiple, now we need to forecast profits per share. To do this, the analysis looks at profits as a proportion of the economy as measured by gross domestic product (GDP). Corporate profits are, at 11 per cent, well above the average level of 8.9 per cent of GDP so Jenkins and Reid assume that, by the end of 2009, the ratio of profit to GDP falls back to its long-term average.

On this basis, assuming that the valuation multiple and profits trend downwards from their present high levels to their long-term average, the Deutsche Bank analysis forecasts returns from US shares of just 1.4 per cent a year in the period 2006-2010. Not enough to pay for the management fee let alone the groceries.

Jenkins and Reid approach the question of future returns from bonds using the same mean-reversion approach: interest rates on 10-year US Government bonds are low given their historical (100-year) relationship with economic growth. They, thus, forecast 10-year bond yields to rise from 4.6 per cent in 2005 to 5.5 per cent in 2010.

When interest rates rise, the capital value of a bond falls and, when some bonds have 30-year maturities, a rise in rates can be sufficient to overwhelm the income return of the bond.

Accordingly, the Deutsche Bank analysis sees US Government bonds returning just 2 per cent a year in 2006-2010 despite the running yield being 4.6 per cent today.

So what does all this mean? First off, Jenkins and Reid's assumption might be wrong. Profits, multiples and interest rates might not mean-revert any time soon and maybe never get back to the 1900-2005 average. It could really be different this time.

But even if mean-reversion doesn't occur, low dividends and interest rates mean that we are probably stuck with low returns for the foreseeable future unless, of course, we have a big crash which lowers the starting point to attractive levels. Remember, too, that the analysis is just for US stocks and, while they represent about 55 per cent of the world, some other major markets are cheaper.

What looks reasonably certain is that individuals counting on 10 per cent returns from the global equity component of their retirement plan look likely to be disappointed unless they have a particularly clever/lucky manager.

The Jenkins/Reid analysis provides an alternative, less optimistic, but more realistic view of long-term future returns from global bond and equity markets than simply assuming historical returns continue.

Their work also highlights the fact that in a world of low single-digit returns annual fee overheads of 2 per cent to 3 per cent a year are ridiculous.

More controversially, however, their computer model is predicting Wales to win the 2007 Rugby World Cup.

In two weeks' time, we take a look at the long-term performance of local bonds and shares and, using data from 1930-2005, ask whether historical returns are realistic estimates for future returns.